Intel Software Optimizations Boost AI Inference in MLPerf v6.1

Intel reports 2.4x higher Llama 3.1 8B server throughput on Xeon 6980P in MLPerf v6.1, achieved via software updates alone. Partner submissions increased from 29 to 39, including first entries from Oracle and Red Hat. These results demonstrate that existing hardware can gain significant inference performance without new silicon purchases.